Supporting discipline under AI Due Diligence
AI Product Due Diligence
Evaluates whether the AI creates real customer and commercial value. A model can work and still be worth nothing if nobody uses it, nobody trusts it, or nobody pays more for it.
Scope
Adoption, value, price
We separate the AI features that appear in marketing from the AI features that appear in usage data and renewal conversations.
Evidence we work from
- Product telemetry by feature, cohort and account size over a trailing period, not a curated window.
- Workflow fit: where the AI sits in the customer's process and what it displaces.
- Trust and review behaviour: override rates, escalation, and whether users check the output.
- Retention and expansion contribution attributable to AI features specifically.
- Willingness to pay: pricing tests, discounting patterns, AI line items in renewals.
Worked example
Marketed versus used
- Technical observation
- Two of five headline AI features are used by under 20% of eligible accounts in a trailing 30-day window, and one of those two is the flagship in all sales collateral.
- Business consequence
- Sales positioning rests on capability the customer base does not actually operate, so renewal conversations turn on features the company under-invests in.
- Investment implication
- Revenue attributed to AI differentiation is overstated. Re-underwrite the expansion case around the proven features and treat activation of the flagship as a value creation milestone with a measurable target.
Headline AI features vs. actual usage
Every feature is marketed to 100% of eligible accounts. Bars show the share that used it in a trailing 30-day window.
Scope
What we assess
The product dimensions we assess — adoption of marketed AI features, workflow fit, and the commercial value the AI actually produces.
- AI
Model architecture
What produces the output, and how much of it the company controls.
- AI
Evaluation discipline
Held-out sets, regression suites, and whether results are reproducible.
- AI
Training & fine-tuning
What was trained, on what, and whether it measurably improved the task.
- AI
Data rights
Licensing, customer terms and training-use permissions behind the corpus.
- AI
Agent reliability
Tool-call errors, escalation rates and observed autonomy.
- AI
Inference economics
Cost per unit of value delivered, and margin at realistic usage.
- AI
Model dependency
Supplier concentration, exit paths and tested fallbacks.
- AI
AI governance
Policy, model change management, human review and audit trail.
- Product
Product adoption
Marketed AI features versus features actually used.
- Product
Workflow fit
Where the AI sits in the customer's process and what it displaces.
Assessing whether the AI is actually worth something?
Bring the thesis, the data room and the timeline. We will tell you what evidence exists, what is missing, and what it means for the deal.